Each year, over 1 million new patients are diagnosed with prostate cancer (PCa) worldwide, and over 300,000 men die of this disease. High risk PCa accounts for the vast majority of PCa deaths. The standard of care for high risk PCa was set by RTOG 92-02, a phase III trial that demonstrated a reduction in disease recurrence and an improvement in PCa-specific survival with radiation therapy (RT) plus long-term androgen deprivation therapy (LTADT, 28 months), compared to RT with short-term ADT (STADT, 4 months). Despite this advance, both overtreatment and undertreatment are acute clinical problems in this patient population. In this trial, 50% of patients were ultimately not cured with RT + LTADT. If this subset of patients had been identified with prognostic biomarkers, they could have received therapy more suited to their aggressive disease, including chemotherapy and novel targeted agents. On the other hand, 30% of men were cured with RT + STADT alone, and could have avoided 24 unnecessary months of exposure to ADT and its toxic side effects. Surprisingly, PCa is one of the few common cancers in which molecular biomarkers are not routinely used to guide therapeutic decisions. To address these unmet needs, we propose to develop and validate clinically useful and cost-effective prognostic and predictive biomarkers for high-risk PCa patients treated with RT, by applying a clinical-grade high-density oligonucleotide array on a unique set of tumor samples from three landmark phase III trials (RTOG 92-02, 99-02, and 94-13). Our research team, combining expertise in PCa, prognostic and predictive biomarker signature identification and validation, bioinformatics, and decision analysis, will: (1) Optimize a prognostic classifier that integrates genomic and clinicopathologic data for PCa patients treated with RT, allowing selection of men with high-risk PCa who would benefit from treatment intensification in future trials, (2) Derive and validate an integrated genomic-clinicopathologic predictor of response to LTADT vs. STADT, a therapeutic duration signature that would allow differentiation of patients who should require LTADT vs. those who are likely to be cured with STADT alone, and (3) Determine the health benefits and cost- effectiveness of using genomic-clinicopathologic classifiers to personalize therapy in men with high-risk PCa. Successful completion of these aims would result in cost-effective prognostic and predictive clinical-grade biomarkers developed on a CLIA-compliant platform that would have an immediate impact on the clinical management of men with high-risk PCa, transforming current treatment paradigms for these patients.